Mon. Not. R. Astron. Soc. 000, 000–000 (2014) Printed 6 May 2016 (MN LATEX style file v2.2) New Lessons from the HI Size-Mass Relation of Galaxies Jing Wang1⋆, B¨arbel S. Koribalski1, Paolo Serra1, Thijs van der Hulst2, Sambit Roychowdhury3, Peter Kamphuis1,4, Jayaram N. Chengalur4 1Australia Telescope National Facility, CSIRO Astronomy and Space Science, PO Box 76, Epping, NSW 1710, Australia 2University of Groningen, Kapteyn Astronomical Institute, Landleven 12, 9747 AD, Groningen, The Netherlands 3Max-Planck Institut f¨ur Astrophysik, D-85748 Garching, Germany 4National Centre for Radio Astrophysics, TIFR, Ganeshkhind, Pune 411007, India Accepted 2014 ???? ?? Received 2014 ???? ??; in original form 2014 January ABSTRACT We revisit the Hi size-mass (DHI-MHI) relation of galaxies with a sample of more than 500 nearby galaxies covering over five orders of magnitude in Hi mass and more than ten B-band magnitudes. The relation is remarkably tight with a scatter σ ∼0.06 dex, or 14%. The scatter does not change as a function of galaxy luminosity, Hi richness or morphological type. The relation is linked to the fact that dwarf and spiral galaxies have a homogenous radial profile of Hi surface density in the outer regions when the radius is normalised by DHI. The early-type disk galaxies typically have shallower Hi radial profiles, indicating a different gas accretion history. We argue that the process of atomic-to-molecular gas conversion or star formation cannot explain the tightness of the DHI-MHI relation. This simple relation puts strong constraints on simulation models for galaxy formation. Key words: intergalactic medium; galaxies 1 INTRODUCTION samples: Verheijen & Sancisi. (2001) for spiral galaxies from the Ursa Major cluster, Swaters et al. (2002) for dwarf and Galaxies are complex ecosystems of gas, stars and dark mat- spiral galaxies, Begum et al. (2008) for dwarf galaxies, Noor- ter governed by the interplay of different processes. Yet, they dermeer et al. (2005) for early-type disk galaxies and Wang may be simpler than expected as many galaxy properties et al. (2013) for massive spiral galaxies. B97 pointed out are well correlated with one another (e.g. Kauffmann et al. that, because the slope is close to 0.5 the relation indicates 2003, Tremonti et al. 2004, Catinella et al. 2010), with mass that the average Hi surface density ΣHI is nearly constant arXiv:1605.01489v1 [astro-ph.GA] 5 May 2016 and environment being probably the most controlling pa- among different types of galaxies. This simple interpretation rameters (Peng et al. 2010). Scaling relations are especially might be the reason why this relation has not been investi- prevalent in Hi-rich, star-forming galaxies as they usually gated further, despite the availability of resolved Hi images have a relatively smooth history for assembling their mass covering a much larger range in M , D and other galaxy (Disney et al. 2008). While the Tully-Fisher relation may be HI HI properties than the individual studies mentioned above. one of the best-known Hi-based scaling relations, galaxies also show tight correlations between their Hi size and mass. This idea is further supported by the finding that galax- The relation between Hi mass (MHI) and the diameter −2 ies have a sharp saturation of ΣHI at ∼ 9 M⊙ pc , where of the Hi disk (DHI) defined at a surface density (ΣHI) of −2 gas at higher surface densities has been converted to molec- 1 M⊙ pc was investigated by Broeils & Rhee (1997, B97 ular gas (Bigiel et al. 2008). However, there is one order of hereafter) and parametrised as: magnitude change in ΣHI between the saturation value and log DHI = 0.51 log MHI − 3.32 (1) where DHI is measured, while the DHI-MHI relation typically has a scatter of less than 0.1 dex. Different galaxies need to −2 where DHI is in unit of kpc and MHI is in unit of M⊙ pc . have an almost uniform distribution function of ΣHI in order Later studies confirmed the DHI-MHI relation (also referred to form a very tight DHI-MHI relation. This is not directly i to as the H size-mass relation in this paper) with other expected because ΣHI is regulated by conversion to molec- ular gas which should vary significantly between galaxies (Leroy et al. 2008). Moreover, it is worth pointing out that ⋆ Email: [email protected] the DHI-MHI relation is not directly linked with the averaged c 2014 RAS 2 Jing Wang et al. ΣHI because a significant fraction of MHI is found outside to exclude these galaxies. For LVHIS, THINGS, LITTLE DHI (this is especially true for early-type galaxies; Serra et THINGS, FIGGS and VIVA, comparisons between inter- al. 2012). ferometric and single dish Hi mass measurements have been In this paper, we will study the scatter and slope of the presented in the relevant reference papers and we only select the DHI-MHI relation using a sample of galaxies with as wide those galaxies where the two Hi mass measurements agree as possible a range in Hi size, mass and other properties. We within 15%. For the WHISP samples we select galaxies with will explore possible explanations for the tightness of the Hi DHI smaller than 400”, as Swaters et al. (2002) estimated mass-size relation by investigating its dependence on other that the missing flux in these galaxies is less than 10% com- parameters. We assume a ΛCDM cosmology with Ωm = 0.3, pared to single-dish fluxes. Galaxies from other samples do Ωlambda = 0.7 and h = 0.7 throughout the paper. not appear to have a missing flux problem. After these selection criteria, there are in total 542 galaxies left (501 unique ones) and they serve as our analy- 2 SAMPLE & DATA sis sample (the sample for Fig. 1, Section 3). We note that the overlapping galaxies have consistent D measurements i HI We compiled H interferometric data from 15 projects and in units of arcsec (the rms scatter of the differences is less summarise the sample names, relevant galaxy numbers, than 0.07 dex). We have access to Hi images for 330 of the types and environment in Table 1. As we will explain be- 542 images (this sub-sample is used in Fig. 4, Section 3). For low, only galaxies with reliable DHI measurements are con- 293 of the 542 galaxies DHI > 3 Bmaj, such that the radial sidered. profile of ΣHI is reasonably resolved (this sub-sample is used We take the values of DHI from published catalogues for in Fig. 2, Section 3). five of the samples. A few galaxies in Ursa Major do not have We retrieve the B-band magnitudes (MB ) and and B- −2 DHI measurements and are excluded. Because the Kovaˇcet band diameter D25 (the major axis for the 25 mag arcsec al. (2009, K09 hereafter, see Table 1) sample has very faint isophote) from the SIMBAD astronomical database2 for 455 systems, we exclude those galaxies with flux uncertainties of the 501 unique galaxies in our sample. We estimate MB larger than 15%. for the Bluedisk galaxies from the g-band magnitudes with We directly measure DHI using the procedure of Wang a correction based on the g − r colour (Jester et al. 2005). et al. (2014) for the other ten samples for which we have We also use g band D to approximate the B-band D for i 25 25 access to the H intensity maps. Whenever possible, we use the Bluedisk galaxies. The g- and r- band data are taken i H images produced with natural weighting in order to have from SDSS (Sloan Digital Sky Survey, York et al. 2000). Ul- high sensitivity to the extended gas. DHI is measured as timately, we are able to obtain optical parameters for 494 i the major axis of a fitted ellipse to the H distribution of the galaxies (the sub-sample for Fig. 3, Section 3). These −2 where the azimuthally averaged ΣHI reaches 1 M⊙ pc . For optical measurements are inhomogeneous, and the uncer- each galaxy from the dwarf and spiral samples, the elliptical tainties are substantial (see, e.g., West et al. (2010) for a shape (position angle and ellipticity) is determined from the discussion of the difficulties associated with measuring op- i H maps, based on the second order moments of the pixel tical magnitudes for extended galaxies such as these). How- −2 distributions where ΣHI > 1 M⊙ pc . For the early-type ever, a full reprocessing of the optical data is beyond the galaxies from Atlas3D we use the elliptical shapes obtained scope of the current work. Therefore, the results based on by Serra et al. (2014) from tilted ring fits to the velocity MB and D25 should be treated with caution. fields. These are more reliable in the case of a disturbed We list the first five galaxies of our full sample along disk morphology, as frequently observed in early type galax- with their Hi and optical parameters investigated in this ies (Serra et al. 2012, 2014). A similar argument applies to paper in Table 2. A full version of the catalogue is available VIVA galaxies in the Virgo Cluster, and the kinematic el- online. liptical shapes are taken from Chung et al. (2009). We note In addition to the main sample, we also collect Hi diam- i that using elliptical shapes determined from H images (as eters and masses for the Milky Way (MW), the Small Mag- we do for the dwarf and spiral galaxies) for the Atlas3D ellanic Cloud (SMC), the Large Magellanic Cloud (LMC), and VIVA galaxies adds some scatter to but does not signif- M31 and a few other special galaxies (Table 3).